[ddots-l] Re: I/O buffer size and latency
- From: "Bryan Smart" <BSmart@xxxxxxxxx>
- To: <ddots-l@xxxxxxxxxxxxx>
- Date: Wed, 18 Apr 2007 22:41:10 -0400
Phil,
First, try running your card at a 96Khz sampling rate. Lower latency is
possible at higher sample rates. That is to say, the lowest latency at
96Khz is lower than the lowest latency at 44Khz. 16 or 24 bits doesn't make
a difference in terms of latency. Once you're at 96Khz, my advice would be
to lower latency as far as is possible before you get drop-outs, and, when
you think that you have a safe setting, back off just a little bit more to
give yourself a little wiggle room. You don't want to always be running
right at the edge, because, while it might work fine most of the time, you
don't want a momentary processing spike to get that extra little atypical
crunch of processing demand that pushes you over into drop-out territory.
Regarding your question about what to expect, with the system that you have,
you should be able to get 2.9Ms latency as an extreme, though, for
stability, I wouldn't push Sonar much more than 5-6Ms. At 5Ms latency, you
should hear only a very short delay.
Bryan
-----Original Message-----
From: ddots-l-bounce@xxxxxxxxxxxxx [mailto:ddots-l-bounce@xxxxxxxxxxxxx] On
Behalf Of Phil Halton
Sent: Wednesday, March 28, 2007 12:33 PM
To: ddots-l@xxxxxxxxxxxxx
Subject: [ddots-l] Re: I/O buffer size and latency
Ok kevin,
the processor is a intel core 2 duo running at 2ghz. the hd is a 7200 rpm.
The problem I am having is:
when I lower the latency slider for my FA66 audio interface (asio drivers),
I begin to get dropouts.
I am recording at 24 bits 44.1khz, and have the asio buffer set to default
settings (normal buffer size and latency slider at 50% ).
When I drop latency slider down to 25%, I begin to get glyches in the
audio--not full dropouts, but stutters and gaps type stuff.
At this setting, latency is quite low (I can hear a minimal lag between
picking a note on my instrument(via mic) and hearing it back through input
echo.
But, naturally I want absolutely no lag between picking and echoing(I
probably ask too much).
I'm just trying to maximize latency for recording acoustic instruments, and
so, I thought maybe the I/O buffer setting might help me squeeze out a
little more latency.
Am I asking too much of software sequencers? Seriously, am I at the limit
of what can be expected from sonar's input echo? Its functional -- which is
more than it ever was with my last system, and I can record in real time
with very little lag. Should I just knock it off and be happy with what I
have?
Finally, If I were to record at 16 bits instead of 24, would that open up
more bandwidth(50% less data with 16 bits compared to 24bits) Since there
would be 50% less data to process I would think it should allow for quicker
response and lower latency settings in the asio buffers.
----- Original Message -----
From: "Kevin L. Gibbs" <kevjazz@xxxxxxx>
To: <ddots-l@xxxxxxxxxxxxx>
Sent: Wednesday, March 28, 2007 5:02 AM
Subject: [ddots-l] Re: I/O buffer size and latency
> 64 is mighty tight. How fast was the processor? What sort of failures do
> you get? Do you get dropouts? If you do, raise the buffer size in
> Options
> Audio.
> K.
>
> -----Original Message-----
> From: ddots-l-bounce@xxxxxxxxxxxxx [mailto:ddots-l-bounce@xxxxxxxxxxxxx]
> On
> Behalf Of Phil Halton
> Sent: Tuesday, March 27, 2007 8:10 PM
> To: ddots-l@xxxxxxxxxxxxx
> Subject: [ddots-l] Re: I/O buffer size and latency
>
>
> So its a guessing game between disk I/O buffer size and soundcard driver
> buffer sizes as to which might cause problems .
>
> Do you think a intel core 2 duo processor with a 7200 rpm HD and 2gb ram
> would benefit from a I/O buffer of 64KB rather than 128?
>
>
> ----- Original Message -----
> From: "Kevin L. Gibbs" <kevjazz@xxxxxxx>
> To: <ddots-l@xxxxxxxxxxxxx>
> Sent: Tuesday, March 27, 2007 10:21 PM
> Subject: [ddots-l] Re: I/O buffer size and latency
>
>
>> The lower the buffer number, the less the latancy. However, if your
>> computer can't cut it, the more dropouts you'll have. If you have a
>> buffer size of, say, 1024, it will be a week between the time you
>> press a key and the time the note sounds.
>> Okay?
>> Kevin
>>
>> -----Original Message-----
>> From: ddots-l-bounce@xxxxxxxxxxxxx
>> [mailto:ddots-l-bounce@xxxxxxxxxxxxx]
>> On
>> Behalf Of Phil Halton
>> Sent: Tuesday, March 27, 2007 5:13 PM
>> To: ddots-l@xxxxxxxxxxxxx
>> Subject: [ddots-l] Re: I/O buffer size and latency
>>
>>
>> Thanks Luis,
>>
>> the fa66 has a faceplate monitor mix knob where I can hear exclusively
>> the output from either the computer, or the inputs, or a mix of both.
>> I kind of
>>
>> like using sonar's input echo because I can wet the signal with reverb
>> or whatever, and hear it exclusively with only a barely noticeable
>> delay by eliminating the dry signal from the mix. But then, I guess
>> I'm just trying to push the limits of this machine and find out what
>> it can do instead of accepting inherant limitations of computer
>> digital audio. I don't know if what I'm after (zero latency input
>> echo monitoring ), is even possible with
>> the current state of computers and software sequencers. So, I might be
>> chasing after rainbows here--so to speak.
>>
>>
>> what you say about I/o buffer size of 64 interests me. Can you tell
>> me
>> why
>> it might be better than 128? I like to understand these things from the
>> bottom up if possible--then I can figure things out for myself once I get
>> the concept. I know this buffer setting is for maximizing disk
>> throughput,
>> but the numbers don't have any rhyme or reason behind them. If you know
>> the
>>
>> why of it, please let me know.
>>
>>
>>
>>
>>
>> ----- Original Message -----
>> From: "Luis Elorza" <luiselorza@xxxxxxxxx>
>> To: <ddots-l@xxxxxxxxxxxxx>
>> Sent: Tuesday, March 27, 2007 7:37 PM
>> Subject: [ddots-l] Re: I/O buffer size and latency
>>
>>
>>>i recommend you use 64 because it will give you enough power to record
>>>a huge project, and i also recommend instead of using input monitor in
>>>your sonar tracks, try using an input monitor in your soundcard, most
>>>have this function that is a direct dry signal routed from your inputs
>>>to your outputs. and this way you can have whatever latency setting in
>>>sonar. any ways you can notice that even at minimum latency your
>>>monitoring will sound as a doubler and to me it's very anoying.
>>> this way you only need low latency when playing soft synths.
>>> ----- Original Message -----
>>> From: "Phil Halton" <philhalt@xxxxxxxxxxx>
>>> To: "ddots-l" <ddots-l@xxxxxxxxxxxxx>
>>> Sent: Tuesday, March 27, 2007 2:06 PM
>>> Subject: [ddots-l] I/O buffer size and latency
>>>
>>>
>>>> my sonar6 producer installation on an ADK laptop is set to a I/O
>>>> buffer size of 128(advanced page of audio options). The sonar
>>>> default is (according to the help files) 64.
>>>>
>>>> I'm using the internal system drive for audio data(7200 rpm drive).
>>>> I'm wondering what the best setting for the I/O buffer size would be
>>>> to maximize latency and minimize dropouts. How does this variable
>>>> affect disk throughput, and with this kind of system drive, what
>>>> might be a better setting than 128KB?
>>>>
>>>> Currently, I can record at 24 bits with minimally perceptable
>>>> latency with input echo on. Naturally, I'm trying for zero latency
>>>> while recording audio, but I start getting dropouts when I start
>>>> cutting the latency slider any further than 50% for the Edirol
>>>> FA66(ASIO). Wouldn't recording at 16 bits improve latency(50%less
>>>> data throughput involved with 16bits from 24bits).
>>>>
>>>> I like the headroom and quality that comes from 24bits, but I want
>>>> that zero latency for obvious reasons. That's why I'm considering
>>>> changing the disk I/O buffer size--it seems to be the only variable
>>>> left that might improve performance.
>>>> .
>>>>
>>>>
>>>> ** To leave the list, click on the immediately-following link:-
>>>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
>>>> ** If this link doesn't work then send a message to:
>>>> ** ddots-l-request@xxxxxxxxxxxxx
>>>> ** and in the Subject line type
>>>> ** unsubscribe
>>>> ** For other list commands such as vacation mode, click on the
>>>> ** immediately-following link:-
>>>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
>>>> ** or send a message, to ** ddots-l-request@xxxxxxxxxxxxx with the
>>>> Subject:- faq
>>>>
>>>
>>> ** To leave the list, click on the immediately-following link:-
>>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
>>> ** If this link doesn't work then send a message to:
>>> ** ddots-l-request@xxxxxxxxxxxxx
>>> ** and in the Subject line type
>>> ** unsubscribe
>>> ** For other list commands such as vacation mode, click on the
>>> ** immediately-following link:-
>>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
>>> ** or send a message, to ** ddots-l-request@xxxxxxxxxxxxx with the
>>> Subject:- faq
>>>
>>>
>>
>> ** To leave the list, click on the immediately-following link:-
>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
>> ** If this link doesn't work then send a message to:
>> ** ddots-l-request@xxxxxxxxxxxxx
>> ** and in the Subject line type
>> ** unsubscribe
>> ** For other list commands such as vacation mode, click on the
>> ** immediately-following link:-
>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
>> ** or send a message, to
>> ** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
>> ** To leave the list, click on the immediately-following link:-
>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
>> ** If this link doesn't work then send a message to:
>> ** ddots-l-request@xxxxxxxxxxxxx
>> ** and in the Subject line type
>> ** unsubscribe
>> ** For other list commands such as vacation mode, click on the
>> ** immediately-following link:-
>> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
>> ** or send a message, to
>> ** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
>>
>>
>
> ** To leave the list, click on the immediately-following link:-
> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
> ** If this link doesn't work then send a message to:
> ** ddots-l-request@xxxxxxxxxxxxx
> ** and in the Subject line type
> ** unsubscribe
> ** For other list commands such as vacation mode, click on the
> ** immediately-following link:-
> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
> ** or send a message, to
> ** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
> ** To leave the list, click on the immediately-following link:-
> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
> ** If this link doesn't work then send a message to:
> ** ddots-l-request@xxxxxxxxxxxxx
> ** and in the Subject line type
> ** unsubscribe
> ** For other list commands such as vacation mode, click on the
> ** immediately-following link:-
> ** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
> ** or send a message, to
> ** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
>
** To leave the list, click on the immediately-following link:-
** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
** If this link doesn't work then send a message to:
** ddots-l-request@xxxxxxxxxxxxx
** and in the Subject line type
** unsubscribe
** For other list commands such as vacation mode, click on the
** immediately-following link:-
** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
** or send a message, to
** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
** To leave the list, click on the immediately-following link:-
** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=unsubscribe]
** If this link doesn't work then send a message to:
** ddots-l-request@xxxxxxxxxxxxx
** and in the Subject line type
** unsubscribe
** For other list commands such as vacation mode, click on the
** immediately-following link:-
** [mailto:ddots-l-request@xxxxxxxxxxxxx?subject=faq]
** or send a message, to
** ddots-l-request@xxxxxxxxxxxxx with the Subject:- faq
Other related posts: